Prognostic factors of drug-resistant epilepsy in childhood: An Italian study

Background: Epilepsy is drug resistant in 30-40% of cases. We studied, retrospectively, the prognostic factors of drug resistance (DR) during a 15 year period, in an Italian sample of patients with childhood epilepsy.

Methods: A total of 117 patients were divided into two groups: one with DR, and the other without DR. The two groups were compared at the following time points: epilepsy onset (T0), and at 2, 5, 8 and 10 years after seizure onset (T2, T5, T8 and T10, respectively) using Fisher's exact test and randomization test. Multiple logistic regression analysis was then used to identify the most reliable predictive model of DR.

Results: Positive neurological examination at onset, symptomatic/probable symptomatic etiology, lack of response to the first drug, seizure clustering during follow up, intelligence quotient ≤ 70, altered neuropsychological examination at onset, and presence of cerebral lesions were predominant in cases of DR. The most reliable combinations of predictors of DR included partial or no response to the first drug, presence of seizure clustering during follow up, altered neurological examination at onset, and long latency between epilepsy onset and first drug at T2; partial or absent response to the first drug and positive magnetic resonance imaging (MRI) at T5; positive MRI and absence of generalized seizures at T8; and positive MRI at T10. DR also sometimes appeared after discontinuation of an effective therapy.

Conclusions: Predictive factors of DR can be recognized in a large number of patients with epilepsy at disease onset, although the current possibility of predicting epilepsy outcome remains limited. In the long term, evidence of cerebral lesions appears to become the most significant prognostic factor.